newt/src/internal/parse.rs

#[derive(Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub struct ValuePath(pub Vec<String>);

pub type Expr = ValuePath;

#[derive(Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub struct If {
	expr: Expr,
	then: Ast,
	else_: Option<Ast>,
}

#[derive(Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub struct For {
	pub expr: Expr,
	pub body: Ast,
}

#[derive(Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub struct Has {
	expr: Box<Expr>,
	body: Ast,
}

#[derive(Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub struct Print(pub Expr);

#[derive(Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
pub enum AstKind {
	Text(String),
	Print(Print),
	Has(Has),
	If(If),
	For(For),
}
pub type Ast = Vec<AstKind>;

pub struct ParseState<'a> {
	template: &'a str,
	last_byte_offset: usize,
	speculative_text_starts_at: usize,
	speculative_text_ends_at: usize,
	current_byte_offset: usize,
	ast: Ast,
}
impl<'a> ParseState<'a> {
	pub fn out_of_input(&self) -> bool {
		self.current_byte_offset >= self.last_byte_offset
	}

	pub fn has_input(&self) -> bool {
		!self.out_of_input()
	}

	pub fn remaining(&self) -> &str {
		&self.template[self.current_byte_offset..]
	}

	pub fn add_current_text_to_ast(&mut self) {
		let text = &self.template
			[self.speculative_text_starts_at..self.speculative_text_ends_at];
		if text.is_empty() {
			return;
		}

		self.ast.push(AstKind::Text(
			self.template[self.speculative_text_starts_at
				..self.speculative_text_ends_at]
				.to_string(),
		));
	}

	pub fn text_might_end(&mut self) {
		self.speculative_text_ends_at = self.current_byte_offset;
	}

	pub fn parse(&mut self) {
		while self.has_input() {
			self.parse_one();
		}

		self.text_might_end();
		self.add_current_text_to_ast();
	}

	pub fn parse_one(&mut self) {
		assert!(self.has_input());
		let started_at = self.current_byte_offset;
		if self.remaining().starts_with('{') {
			self.text_might_end();
			self.maybe_parse_block();
		} else {
			self.current_byte_offset += 1;
		}

		if started_at == self.current_byte_offset {
			panic!("Parser made no progress");
		}
	}

	pub fn maybe_parse_block(&mut self) {
		assert!(self.remaining().starts_with('{'));
		self.current_byte_offset += 1;
		self.skip_whitespace();

		let remaining = self.remaining();
		if remaining.starts_with('.') {
			self.maybe_parse_print();
		} else if remaining.starts_with("for") {
			self.maybe_parse_for();
		}
	}

	pub fn maybe_parse_print(&mut self) {
		assert!(self.remaining().starts_with('.'));
		let path = match self.parse_path() {
			None => return,
			Some(vp) => vp,
		};

		self.skip_whitespace();
		if !self.remaining().starts_with('}') {
			return;
		}
		self.current_byte_offset += 1;

		self.push_node_and_text(AstKind::Print(Print(path)));
		self.new_empty_text();
	}

	pub fn maybe_parse_for(&mut self) {
		assert!(self.remaining().starts_with("for"));
		self.current_byte_offset += "for".len();
		self.skip_whitespace();
		let path = match self.parse_path() {
			None => return,
			Some(vp) => vp,
		};
		self.skip_whitespace();
		if !self.remaining().starts_with('}') {
			return;
		}
		self.current_byte_offset += 1;

		let mut inner = ParseState {
			template: self.template,
			last_byte_offset: self.last_byte_offset,
			speculative_text_starts_at: self.current_byte_offset,
			speculative_text_ends_at: self.current_byte_offset,
			current_byte_offset: self.current_byte_offset,
			ast: Ast::new(),
		};
		while inner.has_input() {
			if inner.remaining().starts_with("{/for}") {
				break;
			}
			inner.parse_one();
		}
		inner.text_might_end();
		inner.add_current_text_to_ast();
		self.current_byte_offset = inner.current_byte_offset;
		if !self.remaining().starts_with("{/for}") {
			return;
		}
		self.current_byte_offset += "{/for}".len();
		let body = inner.ast;

		self.push_node_and_text(AstKind::For(For { expr: path, body }));
		self.new_empty_text();
	}

	pub fn parse_path(&mut self) -> Option<ValuePath> {
		assert!(self.remaining().starts_with('.'));
		self.current_byte_offset += 1;
		let end = self
			.remaining()
			.find(|ch: char| ch == '}' || ch.is_whitespace())
			.unwrap_or(self.last_byte_offset - self.current_byte_offset);
		let inner = &self.remaining()[..end];
		let out = inner.split_terminator('.').map(|s| s.to_string()).collect();
		let out = ValuePath(out);

		self.current_byte_offset += end;

		Some(out)
	}

	pub fn push_node_and_text(&mut self, node: AstKind) {
		self.add_current_text_to_ast();
		self.ast.push(node);
	}

	pub fn new_empty_text(&mut self) {
		self.speculative_text_starts_at = self.current_byte_offset;
		self.speculative_text_ends_at = self.speculative_text_starts_at;
	}

	pub fn skip_whitespace(&mut self) {
		while self.has_input() {
			if self.remaining().starts_with(char::is_whitespace) {
				self.current_byte_offset +=
					self.remaining().chars().next().unwrap().len_utf8();
				continue;
			}
			break;
		}
	}
}

pub fn parse(template: &str) -> Ast {
	let mut state = ParseState {
		template,
		speculative_text_starts_at: 0,
		speculative_text_ends_at: 0,
		current_byte_offset: 0,
		last_byte_offset: template.len(),
		ast: Ast::new(),
	};
	state.parse();

	state.ast
}

#[cfg(test)]
mod test {
	use crate::internal::parse::{AstKind, Print, ValuePath};

	use super::{parse, Ast, For};

	fn go(tmpl: &str, rhs: Ast) {
		let ast = parse(tmpl);
		assert_eq!(ast, rhs);
	}

	#[test]
	fn parse_print_only() {
		go("{.}", vec![AstKind::Print(Print(ValuePath(vec![])))]);
	}

	#[test]
	fn parse_print() {
		go(
			"hello, {.}!",
			vec![
				AstKind::Text("hello, ".to_string()),
				AstKind::Print(Print(ValuePath(vec![]))),
				AstKind::Text("!".to_string()),
			],
		);
	}

	#[test]
	fn parse_print_lookup() {
		go(
			"hello, {.name}!",
			vec![
				AstKind::Text("hello, ".to_string()),
				AstKind::Print(Print(ValuePath(vec!["name".to_string()]))),
				AstKind::Text("!".to_string()),
			],
		);
	}

	#[test]
	fn parse_for() {
		go(
			"{for .}{.}{/for}",
			vec![AstKind::For(For {
				expr: ValuePath(vec![]),
				body: vec![AstKind::Print(Print(ValuePath(vec![])))],
			})],
		);
	}

	#[test]
	fn parse_for_2() {
		go(
			"{for .}{.}, {/for}",
			vec![AstKind::For(For {
				expr: ValuePath(vec![]),
				body: vec![
					AstKind::Print(Print(ValuePath(vec![]))),
					AstKind::Text(", ".to_string()),
				],
			})],
		);
	}

	#[test]
	fn parse_for_nested() {
		go(
			"{for .items}{for .names}{.},{/for}{/for}",
			vec![AstKind::For(For {
				expr: ValuePath(vec!["items".to_string()]),
				body: vec![AstKind::For(For {
					expr: ValuePath(vec!["names".to_string()]),
					body: vec![
						AstKind::Print(Print(ValuePath(vec![]))),
						AstKind::Text(",".to_string()),
					],
				})],
			})],
		)
	}
}